Visible T1-hyperintensity of the dentate nucleus after multiple administrations of macrocyclic gadolinium-based contrast agents: yes or no?

High signal intensity (SI) in the dentate nucleus (DN) and globus pallidus (GP) on unenhanced T1-weighted MR images following the cumulative administration of exclusively macrocyclic gadolinium-based contrast agents (GBCAs) is currently a source of debate and controversy. The widely held view is that visible T1-hyperintensity occurs primarily after the cumulative administration of so-called linear GBCAs rather than macrocyclic GBCAs and that the observed high T1-signal possibly reflects the release of gadolinium (Gd) from the linear chelating molecule and its subsequent retention in the brain, possibly through binding to cellular macromolecules [1‐6]. Although no clinical signs or symptoms attributable to brain Gd retention have yet been observed after cumulative exposure to any GBCA, linear or macrocyclic [7‐11], the widespread assumption is that retained Gd presumed to have been released from linear GBCAs and “visible” on unenhanced T1-weighted images potentially poses a long-term risk to human health [12] while the absence of manifest T1-signal increases with macrocyclic GBCAs is taken as indicating their inherent “safety”. A consequence of this assumption in Europe has been the suspension or restriction of certain linear GBCAs for clinical use [12].

In recent years, numerous studies have demonstrated T1-signal increases in the brain after the exclusive administration of macrocyclic GBCAs [13‐19]. Invariably, however, these studies have been criticized and marginalized, often by the manufacturers of the GBCA in question [20‐23], in part, because the images presented do not show obvious T1-hyperintensity of the DN or GP to the extent seen with linear GBCAs [20]. Unfortunately, such criticism overlooks the fact that brain Gd retention has unequivocally been demonstrated not only in myriad animal studies [24‐31] but also in human brain autopsy samples after the exclusive administration of macrocyclic GBCAs [32].

Patients with relapsing-remitting multiple sclerosis (RR-MS) typically undergo frequent GBCA-enhanced MR studies as part of their routine clinical follow-up. Studies looking at T1-signal changes in the DN and/or GP of patients with RR-MS after the exclusive serial administration of macrocyclic GBCAs have yielded conflicting results. Whereas some studies have failed to observe T1-signal changes [33‐38], others have noted marked quantitative changes [13, 17]. Recently, Splendiani et al. [15] reported visible T1-hyperintensity in the DN in roughly one third of patients with RR-MS who had received cumulative doses of the macrocyclic GBCAs gadobutrol and/or gadoterate. The aim of our study was to further investigate patients with RR-MS for the presence of visible signal enhancement of the DN on unenhanced T1-weighted images following multiple exposures to macrocyclic GBCAs.

A summary of results is presented in Table 1. A total of 149 patients with confirmed RR-MS underwent two or more MRI exams exclusively at our center with gadoterate meglumine (n = 57), gadobutrol (n = 63), or both (n = 29). These 149 patients included 107 (71.8%) for which visible T1-hyperintensity was not evident on any follow-up MRI exam and 42 (28.2%) for which visible T1-hyperintensity was noted. Highly significant (p ≤ 0.0004) quantitative differences in mean DN/P SI ratio were noted for all groups in which visible T1-hyperintensity was noted. Conversely, no significant differences in mean DN/P SI ratio were apparent in any group in which visible T1-hyperintensity was not observed.

The number of administrations per patient in each sub-group is presented in Table 2. A tendency towards a lower number of GBCA administrations per patient was noted for patients with no evidence of visible T1-hyperintensity: 43/107 (40.2%) patients without evidence of visible T1-hyperintensity received five or fewer GBCA administrations compared with only 7/42 (16.7%) patients with evidence of visible T1-hyperintensity. Typically, the first appearance of visible T1-hyperintensity in positive patients occurred at the fourth or fifth follow-up exam, between approximately 3 and 4 years after the initial MRI examination. Few differences were apparent between groups in terms of the time to appearance of the first visible hyperintensity.

Examples of the T1-hyperintensity seen after the exclusive administration of gadoterate meglumine and gadobutrol are shown in Figs. 1 and 2, respectively. The increase in DN/P SI ratio from first to last MRI exam in patients with visible T1-hyperintensity is shown in Figs. 3 and 4, respectively.

Although numerous studies have reported quantitative changes in T1-signal in the DN and/or GP following cumulative administration of macrocyclic GBCAs [13‐19, 41, 42], comparatively few [15, 16, 18, 42] have reported visible T1-hyperintensity. Those studies that have reported visible T1-signal changes following the administration of macrocyclic GBCAs have largely been overlooked or disregarded, even in official guidelines and recommendations [12]. The results of our study provide clear evidence that T1-signal changes can and do occur in the DN after multiple administrations of the macrocyclic GBCAs, gadobutrol and gadoterate meglumine, and that the appearance of visible T1-hyperintensity is associated with significant increases in DN/P SI ratio.

To note is that not all patients that received multiple administrations of macrocyclic GBCAs demonstrated visible T1-hyperintensity. In common with the findings of many previous studies [3‐6], 107/149 (71.8%) patients showed no evidence of visible hyperintensity, even after as many as 20 GBCA administrations. However, it should be noted that patients that demonstrated no signs of visible T1-hyperintensity tended to have received fewer GBCA administrations overall. The appearance of visible T1-hyperintensity in our cohort tended to occur at the 4th or 5th MRI examination whereas approximately 40% of patients negative for visible T1-hyperintensity received five or fewer MRI examinations. It is unclear why some patients demonstrated clear visible T1-hyperintensity of the DN while others did not. Likewise, it is unclear why the relative change in DN/P SI ratio from first to last GBCA administration differed across patients in terms of both magnitude and time course or why the appearance of visible T1-hyperintensity in the DN showed no obvious correlation with quantitative DN/P SI ratio. A recent study by Moreno et al. [18] in melanoma patients referred for serial follow-up gadobutrol-enhanced MRI similarly revealed visible T1-hyperintensity in some, but not all, patients without any association between the DN/P SI ratio and the dose administered or the number of MR examinations. In common with our study, Moreno et al. [18] found a significantly (p = 0.024) higher DN/P SI ratio in patients with visible DN T1-hyperintensity than in patients without visible T1-hyperintensity. Other authors have similarly revealed differences over time in terms of DN/P SI ratio with marked differences across patients in terms of the magnitude of changes [43].

Importantly, none of the patients in our cohort had hepatic deficiency or other clinical issues which might explain a differential appearance of DN T1-hyperintensity and none had had previous cerebral neurosurgical treatments or radiation therapy which might explain the appearance of T1-hyperintensity [39, 40, 44‐46]. Furthermore, there were no relevant differences between SI measurements in the left and right DN of each patient that might suggest a direct association with MS disease progression. On the other hand, MS is a progressive neurodegenerative disease for which patients undergo frequent follow-up MRI examinations and for which disease progression is a potential confounding variable [8]. An early study reported more frequent T1-hyperintensity of the DN in patients with secondary-progressive MS than in patients with primary progressive disease (46% versus 8%, respectively) and found that this correlated with a higher score on the Expanded Disability Status Scale (EDSS), a higher brain lesion load, and greater tissue loss [47]. However, whereas contrast-enhanced T1-weighted images were acquired with either gadoterate or gadodiamide (Omniscan, GE Healthcare) no information was provided on prior GBCA administrations, so it is difficult to determine whether the observed T1-hyperintensity reflected differences in the stage/extent of disease or differences in the amount/type of GBCA administered. Although some studies [48, 49] have looked to correlate T1-hyperintensity of the DN in patients with MS with loss of verbal fluency and cognition, it is difficult to separate a hypothetic effect of Gd retention from the normal effects of MS progression, especially given the difficulties in finding a matched MS group not exposed to any GBCA. Recent studies by Cocozza et al. [9] and Ackermans et al. [10] suggest that MRI features suggestive of Gd retention do not correlate with EDSS worsening in patients with MS. These findings are supported by studies that have shown no effects of DN T1-hyperintensity on DN tissue integrity [11, 50, 51].

Our findings support those of Stojanov et al. [13] and Splendiani et al. [15] in showing significant increases in DN/P SI ratio following the cumulative administration of macrocyclic GBCAs in patients with RR-MS. Conversely, they are at variance with the findings of others [33‐38]. The clear disparity in study findings between research groups highlights the complexity of the issues at hand and the need for carefully controlled large-scale studies to investigate the processes underlying T1-signal changes in these patients. Studies are needed to ascertain whether T1-hyperintensity in the DN is in fact indicative of Gd retention or whether it represents some other process specific to MS and/or other diseases. In the absence of histologic verification of Gd retention, no conclusions can be made as to the processes underlying the T1-signal changes observed in our patient cohort. However, this is the same for all studies that focus solely on observations from imaging studies.

The strength of our study lies in the fact that all patients received only macrocyclic GBCAs and that imaging was performed in a carefully monitored manner exclusively at our center. Therefore, the possibility that patients with visible T1-hyperintensity could have received prior administrations of certain linear GBCAs that are more widely associated with T1-signal changes in the DN can be excluded. Secondly, images from all patients that received at least two macrocyclic GBCA administrations were carefully assessed for visible T1-hyperintensity by two neuroradiologists in consensus. Prior studies to evaluate T1-signal changes have frequently limited the number of patients to a convenient round number (e.g., n = 50 [6]), restricted patient inclusion to only those that received a minimum number of GBCA administrations (e.g., n = 6), and cited SI values determined by just a solitary reader. Our approach avoided any form of selection bias and was based solely on the visual appearance of the DN relative to the pons. Our approach therefore more typically reflects the assessment process routinely undertaken in clinical practice. Finally, our study included patients with RR-MS evaluated exclusively at our center over an extended period (approximately 13 years). Other studies have drawn conclusions based on very few evaluated patients imaged over a relatively short time frame [37].

A limitation of our study is the absence of an age-matched control group of MS patients that had never been exposed to any GBCA. However, such patients would be extremely difficult to find given that contrast-enhanced MRI is one of the principal imaging techniques for MS patients, both for initial diagnosis and follow-up. A second limitation is that we did not quantify the DN/P SI ratio at every follow-up exam and thus were unable to detail a time course for T1-signal changes in individual patients. However, in quantifying changes in DN/P SI ratio from SI measurements determined prior to the first and last contrast-enhanced exam in each patient, we showed that increases in DN/P SI ratio can and do occur after both gadoterate and gadobutrol and that the changes in T1-signal are variable across patients both in terms of magnitude and time course. Unfortunately, a statistical comparison of these two GBCAs in terms of the changes observed was beyond the scope of this initial study given the relatively small number of positive patients in each group. Finally, we did not attempt to correlate the appearance of visible T1-hyperintensity with cognitive or physical disability as performed elsewhere [9, 10, 49]. Moreover, we did not look at a possible correlation with brain lesion load and tissue loss. This was considered beyond the scope of this initial investigation but should be considered for further large-scale studies in MS patients. Further research should additionally look at changes in R1 relaxation rates which are less prone to potential sources of inaccuracy.

In conclusion, we observed visible T1-hyperintensity of the DN in 42/149 (28.2%) patients with RR-MS who had received at least two administrations of gadobutrol, gadoterate, or both. The appearance of visible T1-hyperintensity correlated with significant increases in DN/P SI ratio from first to last examination.